Hey!
Apologies if this isnt the right place to ask this question, as it is a theoretical question based on the paper, and not the implementation!
To generate the dataset, Dex-Net 2 introduces Eq - the robust epsilon quality. The success metric S is 1 if this Eq > δ .
Dex-Net 1 introduces Probability of Force closure, where occurence of a force closure is 1 if the epsilon quality > δ i.e. convex hull
(Please correct me if my understanding is wrong )
So, what is the difference between the robust epsilon quality introduced in DexNet2 and the probability of force closure in Dex-Net1.0
In a slightly broader sense, how different is DexNet2 from DexNet1 (apart from the GQCNN, which if I understand correctly is a standalone module on its own)?
Hey! Apologies if this isnt the right place to ask this question, as it is a theoretical question based on the paper, and not the implementation!
To generate the dataset, Dex-Net 2 introduces Eq - the robust epsilon quality. The success metric S is 1 if this Eq > δ . Dex-Net 1 introduces Probability of Force closure, where occurence of a force closure is 1 if the epsilon quality > δ i.e. convex hull (Please correct me if my understanding is wrong )
So, what is the difference between the robust epsilon quality introduced in DexNet2 and the probability of force closure in Dex-Net1.0
In a slightly broader sense, how different is DexNet2 from DexNet1 (apart from the GQCNN, which if I understand correctly is a standalone module on its own)?